The impact of histological reclassification during pathology re-review--evidence of a Will Rogers effect in bladder cancer?

Purpose: We investigated the association of histological reclassification during pathology re-review of radical cystectomy specimens with clinicopathological outcomes in patients initially classified with urothelial carcinoma.

Materials and methods: We identified 1,211 patients initially diagnosed with urothelial carcinoma at radical cystectomy between 1980 and 2005. All pathological specimens were re-reviewed by a urological pathologist. Survival was estimated using the Kaplan-Meier method and compared with the log rank test.

Results: Of 1,211 cases previously recorded as pure urothelial carcinoma 406 (33%) were reclassified as variant histology. The most common variant histologies identified were squamous in 151 patients (37%) and micropapillary in 62 (15%). Variant histology on re-review was associated with a higher rate of extravesical disease (71%) than urothelial carcinoma at initial diagnosis (52%) or pure urothelial carcinoma on re-review (42%, p<0.0001). Median postoperative followup was 11.1 years, during which 976 patients died, including 564 of bladder cancer. Notably, reclassification resulted in significant stratification of 10-year cancer specific survival, which was 50% in patients with pure urothelial carcinoma after re-review, 47% in those with urothelial carcinoma on initial interpretation and 42% in those with variant histology on re-review (p=0.03). Ten-year overall survival in patients with urothelial carcinoma on re-review, urothelial carcinoma at initial interpretation and variant histology on re-review was 29%, 27% and 24%, respectively (p=0.04).

Conclusions: Pathological re-review of radical cystectomy specimens identified variant histology in a third of patients. These variants are associated with a high rate of locally advanced disease, which may impact the noted rates of cancer specific and overall survival. Thus, it is critical to be aware of re-review status when interpreting outcomes from historical data sets and stratifying risk.